On 14.01.2011 09:04, Stefan Kaitschick wrote: > Something along the line of finding move equivalency classes by > statistical means. This is yet another possible approach. It does not refute my idea though. -- robert jasiek